Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
Salmonella enterica serovar Typhimurium is capable of producing cellulose as the main exopolysaccharide compound of the biofilm matrix. It has been shown for Gluconacetobacter xylinum that cellulose biosynthesis is allosterically regulated by bis-(3′,5′) cyclic diguanylic acid, whose synthesis/degradation depends on diguanylate cyclase/phosphodiesterase enzymatic activities. A protein domain, named GGDEF, is present in all diguanylate cyclase/phosphodiesterase enzymes that have been studied to date. In this study, we analysed the molecular mechanisms responsible for the failure of Salmonella typhimurium strain SL1344 to form biofilms under different environmental conditions. Using a complementation assay, we were able to identify two genes, which can restore the biofilm defect of SL1344 when expressed from the plasmid pBR328. Based on the observation that one of the genes, STM1987, contains a GGDEF domain, and the other, mlrA, indirectly controls the expression of another GGDEF protein, AdrA, we proceeded on a mutational analysis of the additional GG[DE]EF motif containing proteins of S. typhimurium. Our results demonstrated that MlrA, and thus AdrA, is required for cellulose production and biofilm formation in LB complex medium whereas STM1987 (GGDEF domain containing protein A, gcpA) is critical for biofilm formation in the nutrient-deficient medium, ATM. Insertional inactivation of the other six members of the GGDEF family (gcpB-G) showed that only deletion of yciR (gcpE) affected cellulose production and biofilm formation. However, when provided on plasmid pBR328, most of the members of the GGDEF family showed a strong dominant phenotype able to bypass the need for AdrA and GcpA respectively. Altogether, these results indicate that most GGDEF proteins of S. typhimurium are functionally related, probably by controlling the levels of the same final product (cyclic di-GMP), which include among its regulatory targets the cellulose production and biofilm formation of S. typhimurium.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Biofilm formation capacity appears to be widespread among natural isolates of Salmonella enteritidis and Salmonella typhimurium (Zogaj et al., 2001; Solano et al., 2002) and benefits the survival of Salmonella on environmental surface habitats (Solano et al., 2002). These organisms form diverse types of biofilms depending on environmental conditions (Romling et al., 1998a; Romling and Rohde, 1999; Solano et al., 2002). Thus, the biofilm produced at room temperature in a rich medium, LB, is made of several components among which two, fimbriae and cellulose, have been well characterized. Their expression leads to the formation of a pellicle containing a tight bacterial network on the air–broth interface (Zogaj et al., 2001; Solano et al., 2002). On the other hand, cellulose seems to be the main component of the biofilm formed when bacteria are grown at 37°C in a nutrient-deficient medium, ATM (Solano et al., 2002).
Genes involved in cellulose biosynthesis in S. enteritidis and S. typhimurium are organized in two divergently transcribed operons, bcsABZC and bcsEFG (Zogaj et al., 2001; Solano et al., 2002). The presence of bcs operons is not restricted to Salmonella and homologous gene clusters are present in all the species investigated of the family Enterobacteriaceae (Zogaj et al., 2003; our unpublished results). Homology searches have shown that bcsA, bcsB and bcsZ encode for the catalytic subunit of the cellulose synthase, the regulatory subunit that binds to the activator cyclic diguanylic acid, c-di-GMP, and an endoglucanase respectively. In the second operon, bcsG also shows homology to an endoglucanase gene. bcs genes appear to be constitutively transcribed and consequently their transcription is not dependent on CsgD or AdrA (Zogaj et al., 2001). Apart from this, very little is known about the biosynthesis of cellulose in these bacteria.
The model organism for the characterization of regulatory elements involved in cellulose production has been Gluconacetobacter xylinus[formerly known as Acetobacter xylinus(m)]. In this bacterium, glucose units in UDPG (Uridin-5′-diphosphoglucose) are incorporated into the cellulose polymer (1,4-β-D-glucan) by a membrane-bound cellulose synthase, which is specifically dependent on c-di-GMP (Ross et al., 1987; 1990; Weinhouse et al., 1997). The cellulose synthesis is therefore modulated by the opposing action of two enzymes, diguanylate cyclase (DGC) and c-di-GMP diesterase (PDEA), controlling the level of c-di-GMP in the cell (for review, see Ross et al., 1991). After an extensive study, Tal et al. (1998) characterized the dgc and pdeA genes as controlling the turnover of this novel effector. The identification of two adjacent domains designated as GGDEF and EAL shared by all the DGC and PDEA isoenzymes suggested that either or both domains were involved in its diguanylate cyclase activity (Ausmees et al., 2001). Remarkably, the GGDEF domain is very abundant in the genomes of free-living bacteria but most of the GGDEF proteins have not yet been experimentally characterized (Galperin et al., 2001).
To get further insight into the regulation of cellulose synthesis and biofilm formation in Salmonella, and taking advantage of the properties of this organism, this study used two mouse virulent strains of S. typhimurium, ATCC14028s (Fields et al., 1986) and SL1344 (Hoiseth and Stocker, 1981), that differ in their abilities to form a biofilm. Two plasmids from a genomic bank of the biofilm-forming strain S. typhimurium ATCC14028s bestowed strain SL1344 with the capacity for cellulose synthesis and biofilm production. As a result, in this report, we identify a novel protein, STM1987, containing a GGDEF domain and possibly possessing diguanylate cyclase activity that confers strain SL1344 with the capacity for cellulose biosynthesis and biofilm formation in ATM media. This activity exhibited a dependence on the presence of the conserved sequence pattern GG[DE]EF motif present in the GGDEF domain. Also, by carrying out non-polar mutation and complementation experiments of either STM1987 or adrA mutants, we investigated the role of all S. typhimurium proteins containing the conserved GG[DE]EF motif on cellulose biosynthesis and biofilm formation. As a result, we identified five new members of the GGDEF family involved in cellulose biosynthesis by S. typhimurium. We named these proteins as Gcp (GGDEF domain-containing protein). According to our results, the GGDEF protein family in Salmonella appears to connect different environmental signals to cellulose production and biofilm formation and to fulfil diverse functions in the cell.
Salmonella typhimurium 14028 DNA bestows S. typhimurium SL1344 with the capacity for biofilm formation
In a previous study where we analysed the biofilm formation capacity of S. enteritidis and S. typhimurium strains (Solano et al., 2002), we observed that strain 14028 was capable of biofilm formation both in a rich medium (LB) at room temperature and in a nutrient-deficient medium (ATM) at 37°C. Also, this strain fully fluoresced on calcofluor agar plates, which indicated the cellulose-producing capacity of this strain. On the other hand, strain SL1344 did not form a biofilm under similar conditions (LB and ATM) and did not fluoresce on calcofluor agar plates (Fig. 1A). To determine why cellulose production was impaired in S. typhimurium SL1344, we complemented it with a genomic DNA library of strain 14028 constructed in pBR328. A collection of 30 000 chloramphenicol-resistant and tetracycline-sensitive clones were screened for their ability to bind calcofluor on agar plates. Sixty-two fluorescent clones were selected and further analysed for their biofilm formation capacity in LB and ATM conditions. As a result, we found that 46 (group I) out of the 62 fluorescent selected clones had gained the capacity to form a biofilm in ATM but not in LB (Fig. 1B–D). In contrast, the remaining 16 selected clones (group II) had gained the capacity to form a biofilm in LB but not in ATM (Fig. 1B–D). To exclude fortuitous problems, such as spontaneous mutations, each plasmid was backcrossed into the wild-type strain SL1344 by electroporation or P22 transduction and phenotypes were confirmed. These results indicated that strain SL1344 either is deficient in genes or presents a deficit in the expression of genes essential for cellulose biosynthesis and biofilm formation in different environmental conditions.
STM1987 and MlrA activate cellulose production and biofilm formation in S. typhimurium
To identify the genes whose expression bestows SL1344 strain with the capacity for cellulose biosynthesis and biofilm formation, the flanking regions of the inserts carried by plasmids extracted from fluorescent complemented strains were sequenced. Homology searches using the blast program at the NCBI server showed that all the clones of group I contained a plasmid, designated pI, which carried a DNA region of the S. typhimurium LT2 genome extending from 2068681pb to 2075357pb (Fig. 2). On the other hand, all the clones of group II contained a plasmid, pII, which carried a DNA region extending from 2252246pb to 2257964pb of the S. typhimurium LT2 genome (Fig. 2). Thus, two independent inserts were capable of bestowing biofilm formation capacity to strain SL1344 in different environmental conditions. To determine the genes of both inserts responsible for the switching on of cellulose synthesis and biofilm formation in strain SL1344, each ORF was subcloned in pBR328 and introduced into strain SL1344. The subclone pIb, which contained an open reading frame designated STM1987, bestowed strain SL1344 with an identical phenotype to that obtained with the complementation with the pI plasmid. The region of the genome carried by pII included the mlrA gene, the product of which has been demonstrated to positively regulate curly synthesis and extracellular matrix formation in Escherichia coli and S. typhimurium (Brown et al., 2001). Consistent with these results, the subclone pIIc containing the mlrA gene bestowed SL1344 strain with the capacity to form a rigid pellicle on LB and a calcofluor fluorescence phenotype.
To confirm the role of STM1987 and mlrA in biofilm production by S. typhimurium, non-polar mutations of STM1987 and mlrA were constructed by one-step inactivation in strain 14028 (see Experimental procedures). The resulting strain 14028 ΔSTM1987-Km lacked the capacity to produce a biofilm in ATM but maintained its ability to produce a biofilm under LB conditions and to bind calcofluor on agar plates. On the other hand, 14028 ΔmlrA-Apr lacked the capacity to fluoresce on calcofluor plates or form a biofilm in LB but maintained its ability to form a biofilm in ATM (Fig. 3). Moreover, STM1987 and mlrA mutations were moved from S. typhimurium 14028 to the biofilm-forming S. enteritidis strain 3934 via P22 transduction. The resulting mutant strains, verified by polymerase chain reaction (PCR), showed the same phenotypes as 14028 ΔSTM1987-Km and 14028 ΔmlrA-Apr (data not shown).
Altogether, these results showed that STM1987 is required for cellulose production and biofilm formation in ATM and not for biofilm formation in LB conditions or binding to calcofluor on agar plates. Intriguingly, as stated above, we had identified STM1987 by complementation and restoration of calcofluor fluorescence of strain SL1344. Contrary to this, a mutation of the STM1987 gene in a biofilm-producing strain, such as S. typhimurium 14028, did not affect the cellulose production capacity. Such results suggested that under LB incubation conditions, another genetic component different from STM1987, probably regulated by mlrA, is responsible for the cellulose-producing capacity of strain 14028. On the other hand, mlrA showed not to be needed for biofilm formation in ATM.
MlrA transcription is defective in strain SL1344
The mlrA gene has been identified as a positive regulator of the csgD gene, aggregative morphology and extracellular matrix formation in E. coli and S. typhimurium (Brown et al., 2001). S. typhimurium SL1344 contains the mlrA gene, as verified by PCR, but it is incapable of biofilm formation. The fact that strain SL1344 recuperated its capacity to form a biofilm in LB when complemented with the mlrA gene of strain 14028 could suggest either that the mlrA gene of SL1344 is not correctly transcribed or that the MlrA protein produced by strain SL1344 is not functional. To account for this, the wild-type SL1344 strain was complemented with the mlrA gene amplified from SL1344 and cloned in the pBR328 plasmid. Complementation of this strain resulted in the restoration of cellulose production and the biofilm-forming phenotype in LB. These results suggested that the mlrA gene product from strain SL1344 was a functional protein and it therefore seemed reasonable to suppose that a decreased transcription level could be responsible for the biofilm-deficient phenotype of strain SL1344. To confirm this hypothesis, we used real-time quantitative PCR. The results showed that the level of expression of the mlrA gene in strain SL1344 was significantly decreased (P < 0.05) compared with the expression of mlrA in strain 14028 (Fig. 4A).
As stated before, MlrA is known to control the transcription of the response regulator CsgD, which in turn regulates the biosynthesis of thin aggregative fimbriae and the transcription of adrA. Both fimbriae and cellulose are main components of the biofilm formed by Salmonella in LB. Accordingly, the cellulose synthesis and biofilm formation deficiency of strain SL1344 could result from an insufficient transcriptional activation of csgD and be in consequence of adrA. To test this assumption, wild-type strain SL1344 was complemented with the adrA and csgD genes cloned in a low-copy plasmid (pBR328). Complementation with AdrA resulted in the restoration of cellulose synthesis, as shown by calcofluor fluorescence on agar plates but the strain remained incapable of biofilm formation. The wild-type strain SL1344 complemented with csgD gained the capacity not only to fluoresce on calcofluor agar plates but also to form a biofilm in LB conditions as a result of cellulose and fimbriae biosynthesis (data not shown). Reverse transcriptase polymerase chain reaction (RT-PCR) experiments were carried out to establish the possible transcriptional relationship between mlrA and adrA. Total RNA of the wild-type strain 1344 and the complemented strain with mlrA was isolated after 24, 48 and 72 h of incubation in LB under biofilm-forming conditions. The expression of adrA was found to be significantly increased in the complemented strain with respect to the wild-type strain and to be at its maximum expressiveness at 72 h of incubation during the formation of a biofilm in LB (Fig. 4B). Overall, these results show that the mlrA transcription deficiency in strain SL1344 provokes a deficit in the levels of the AdrA protein in the cell that leads to the absence of cellulose synthesis and the biofilm-negative phenotype in LB.
STM1987 activates cellulose production of an adrA mutant
The 1713 bp STM1987 gene is predicted to encode a protein of 570 amino acids and 65.42 kDa. Homology searches of the STM1987 gene product established that STM1987 shows a high identity (overall identities > 65%) to hypothetical proteins present in Shigella flexneri (AE015217), E. coli O157:H7 (BAB36117), E. coli K12 (AAC75022), E. coli CFT073 (AE016762) and Salmonella typhi (CAD05734). Domain analysis of the STM1987 protein using the tmhmm prediction program (tmhmm Server V. 2.0; CBS) and the Pfam database showed the presence of two putative transmembrane helices (aa 20–42, 358–380) and a GGDEF domain at the C-terminal region of the protein (aa 394–557). Because of the presence of the GGDEF domain, the STM1987 protein presents significant alignments to GGDEF proteins for which a function has been already proposed (TableS1 in Supplementary material). Notably, most of these proteins have been found to be involved in the regulation of polysaccharide production and/or in the biofilm formation process.
Tal et al. (1998) reported the isolation of the cdg1, cdg2 and cdg3 operons, which encode isoforms of DGC and PDEA that hierarchically contribute to cellulose production in A. xylinum. Also, genetic complementation using genes of different bacteria encoding proteins with a GGDEF domain as the only element in common restored cellulose synthesis in a cellulose-deficient Rhizobium leguminosarum bv. trifolii strain (Ausmees et al., 2001). It might be possible that in Salmonella, proteins sharing the GGDEF domain could complement each other and accomplish the same function in the cell. Interestingly, AdrA contains a GGDEF domain (Romling et al., 2000). To test whether AdrA and STM1987 are functionally related, we constructed an insertional adrA mutant that was used together with an STM1987 mutant in epistasis experiments. As reported earlier (Romling et al., 2000), the resulting adrA mutant strain was defective in the calcofluor-binding activity and formed a fragile pellicle at the air–liquid interface under LB biofilm-forming conditions. On the other hand, in ATM biofilm-forming conditions, the adrA mutant strain behaved like the wild-type strain. The adrA mutant strain was then complemented with the STM1987 gene expressed in the pBR328 plasmid and the resulting complemented strain regained the capacity to form a biofilm in LB and to fluoresce on calcofluor agar plates (Fig. 5A). However, complementation of the STM1987 mutant strain with adrA expressed in the pBR328 plasmid did not lead to the restoration of the biofilm-forming phenotype in ATM (Fig. 5B), although RT-PCR experiments confirmed that the transcriptional level of adrA was significantly increased in the complemented strain with respect to the wild-type strain (data not shown). The correspondent complementation controls were carried out, namely complementation of the adrA and STM1987 mutant with adrA and STM1987 respectively. Both complemented strains recuperated their corresponding phenotypes in LB and ATM (Fig. 5). Altogether, these data indicate that first, in the absence of AdrA, STM1987 is capable of activating cellulose synthesis and biofilm formation in LB, and second, AdrA is not capable of assuming the role of STM1987 in ATM media.
The GGEEF motif in the GGDEF domain is essential for the activity of STM1987
The GGDEF domain is present in all proteins known to be involved in the regulation of cellulose synthesis in many bacteria (TableS1). Among the consensus sequence of the GGDEF domain, the highly conserved five-amino-acid residues motif GG[DE]EF is present (Galperin et al., 2001). To obtain evidence of whether the GGEEF motif present in the GGDEF domain of STM1987 mediated the protein activity, acidic amino acid residues corresponding to positions 3 and 4 of the consensus motif were mutated to Gly and Ser. Thus, we obtained a recombinant STM1987 gene (STM1987-GGGSF) that contained the following amino acid substitutions: E479G and E480S. Complementation of the STM1987 mutant strain with the STM1987-GGGSF gene expressed in pBR328 did not lead to the restoration of the biofilm-forming phenotype in ATM, indicating that an intact GGEEF motif is essential for the function of the protein (data not shown).
Role of GGDEF proteins in biofilm formation in Salmonella
Eleven proteins containing the GGDEF domain have been annotated as putative diguanylate cyclases/phosphodiesterases on the complete genome of S. typhimurium (Genome Sequencing Center, Washington University School of Medicine, MO, USA) (Table 1). Apart from AdrA and STM1987, none of the remaining nine GGDEF proteins have yet been characterized, strongly suggesting that we are missing major aspects of the functions related to these proteins. Three of these GGDEF proteins, namely YhdA, YfeA and YhjK, do not share the consensus GG[DE]EF motif and consequently were not taken into consideration. STM1987 and the other six novel proteins that share the GGDEF domain and contain the consensus GG[DE]EF motif were further studied and designated as Gcp (GGDEF domain containing protein) (Table 1). Sequence alignments revealed that GcpA (STM1987) is one of the simplest members of the GGDEF family in S. typhimurium, consisting of only one known domain. To establish whether Gcp proteins play a role in biofilm production, isogenic derivatives of S. typhimurium 14028 were constructed by one-step insertional mutation. The resulting mutant strains maintained their calcofluor-binding properties as well as their biofilm-forming capacities in LB and in ATM (Fig. 6). Notably, the S. typhimurium 14028 gcpE mutant strain displayed an increased brightness on calcofluor agar plates. The gcpA and adrA mutant strains were then systematically complemented with each of the gcp genes expressed in pBR328. RT-PCR experiments were performed to confirm the increased transcriptional level of each gene in the complemented strain with respect to the wild-type strain. The resulting adrA mutant complemented strain with gcpA, B, C and F regained the capacity to form a biofilm in LB and fluoresce on calcofluor agar plates as achieved by complementation with the gcpA gene (Fig. 7A). On the other hand, GcpB, C, D and F were also able to activate cellulose synthesis of a gcpA-defective strain under ATM conditions (Fig. 7B). In agreement with the mutation studies, complementation of S. typhimurium 14028 wild-type strain, adrA and gcpA mutant strains with the gcpE gene totally abolished biofilm formation and cellulose synthesis in calcofluor and LB (Fig. 8, see also Fig. 7A and B). Overall, the results indicated that different proteins, namely GcpA, B, C and F, are capable of assuming the role of AdrA in LB conditions and that GcpB, C, D and F complement the GcpA activity in ATM. Only GcpG was unable to restore cellulose synthesis in any of the conditions tested. Also, GcpE appears to be a suppressor of cellulose synthesis under LB conditions, probably by degrading the allosteric effector c-di-GMP.
Table 1. Proteins of the Salmonella genome that contain the GGDEF domain according to the Pfam program.
Real-time quantitative PCR to investigate the relationship between the transcriptional activator mlrA and the Gcp proteins in S. typhimurium was used. Total RNA of the wild-type strain SL1344 and the complemented strain with mlrA was isolated at 24, 48 and 72 h of incubation in LB under biofilm-forming conditions and the level of expression of gcp genes was studied. The level of csgD expression, measured as a control, was significantly increased (P < 0.05) in the complemented strain. However, the increased transcription of csgD displayed by the complemented strain with mlrA was found not to affect the expression of any of the gcp genes (Fig. 9). Based on these results, we concluded that under the conditions tested neither mlrA nor csgD affect transcription of gcpA, B, C, D, E, F and G.
We used a complementation strategy in an S. typhimurium strain incapable of producing cellulose to identify new genes required for the cellulose synthesis process. S. typhimurium SL1344, a strain routinely used in many laboratories for virulence studies that as a consequence of domestication has lost the ability to synthesize cellulose, was chosen. It is worth pointing out that strain SL1344 was used by Brown et al. (2001) to determine the role of mlrA in curli production and extracellular matrix formation, which indicates that isolates of this strain from other laboratories might be able to produce cellulose and a biofilm. In this study, the first unexpected finding was that two different genes, mlrA and gcpA (STM1987), were able to bestow cellulose production capacity to the same strain. Our results demonstrate that the link between both genes is actually the presence of a GGDEF domain in both GcpA as well as an inner surface protein known as AdrA, which is regulated by mlrA. The SL1344 strain showed a deficiency in the transcription of mlrA, which is a transcriptional activator of csgD in S. typhimurium and E. coli (Brown et al., 2001). The csgD gene product co-regulates the synthesis of both fimbriae and AdrA, a protein essential for cellulose biosynthesis (Romling et al., 2000; Gerstel et al., 2003). The reasons why transcription of mlrA is deficient in strain SL1344 remain unclear. MlrA expression is positively regulated by RpoS and therefore a deficient expression of rpoS could be responsible for the cellulose-deficient phenotype of strain SL1344 (Brown et al., 2001). However, SL1344 exhibited similar levels of catalase activity, whose expression is activated by RpoS, to other biofilm-positive wild-type strains. Furthermore, complementation experiments of SL1344 with rpoS restored neither cellulose production nor biofilm formation, indicating that the overexpression of rpoS could not restore the mlrA levels in strain SL1344 (our unpublished results).
GcpA required to be overexpressed to bypass the AdrA defect caused by deletion of the adrA gene in strain 14028. One possible explanation for the inability of normal expression levels of GcpA to assume the role of AdrA in its absence could be that GcpA may not be as efficient as AdrA in LB conditions and therefore, high levels of the GcpA protein may be required to overcome the AdrA deficiency. Remarkably, overexpression of adrA in ATM conditions could not bypass a GcpA defect. This finding indicates that AdrA is not active at 37°C, in minimal medium under strong shaking conditions. An extensive study of environmental conditions that influence csgD expression and cellulose production (regarded as the distinctive rdar morphotype on agar plates) performed by the group of U. Romling showed that nutrient depletion, microaerophilic conditions in rich medium and aerobic conditions in minimal medium, alkaline pH (pH 8.5), low osmolarity and a temperature of 28°C activated the transcription of csgD and the ensuing cellulose overproduction (Gerstel and Romling, 2001; Gerstel et al., 2003). In addition to the highly refined regulation of csgD expression, our results suggest that cellulose synthesis regulation could take place at the level of members of the GGDEF family involved in the cellulose synthesis process.
A common striking feature between AdrA and GcpA is that both contain a GGDEF domain. The presence in the genome of S. typhimurium of another six proteins of unknown function containing the GGDEF domain with the conserved sequence GG[DE]EF motif, which we have proven to be essential for the functionality of GcpA, encouraged us to investigate further as to whether other members of the GGDEF family could affect the cellulose biosynthesis pathway. Interestingly, none of the gcpB to gcpG mutants showed any deficiency in cellulose production and biofilm formation either in LB or in ATM conditions. These results showed that the presence of AdrA or GcpA overcome the absence of the other members of the family. In contrast, complementation experiments of both adrA and gcpA mutants with each gcp gene showed that overexpression of gcpB, C and F restored cellulose synthesis and biofilm formation in both environmental conditions. GcpD could only complement GcpA deficiency in ATM whereas overexpression of gcpG was unable to restore cellulose synthesis in any of the conditions tested. Interestingly, gcpG is the only member of the GGDEF family of S. typhimurium that takes part of an operon with yfiR, a putative periplasmic protein of unknown function. The possibility that GcpG requires the presence of YfiR for its activity cannot be ruled out. Another peculiarity of GcpG is that it harbours the HAMP signal transduction domain. This domain has been found in many transmembrane receptors and it has been proposed to be involved in dimerization process (Aravind and Ponting, 1999; Zhulin et al., 2003). It is tempting to speculate about the existence of an order or hierarchy in signal transduction between both domains that hinder the GGDEF domain functionality in our experimental conditions.
Overall, our results suggest that AdrA, GcpA, B, C, D and F are able to catalyse the same enzymatic activity, which renders a common final product, c-di-GMP. We cannot exclude the possibility that the main function of all these proteins could be the tight regulation of cellulose biosynthesis and biofilm formation under different environmental conditions. However, it appears that these proteins are more likely to be involved in signal transduction of different metabolic pathways, and the fact that they share the final product of their enzymatic activity makes them able to bypass the absence of other members of the family. There is abundant evidence in support of this hypothesis. First, overexpression of gcp genes under their own promoters in ΔadrA and ΔgcpA strains showed that GcpA and GcpB repressed swarming and swimming motility (data not shown), indicating that other regulatory circuits different to cellulose biosynthesis can be indirectly affected when GGDEF proteins are overexpressed. A relationship between GGDEF proteins and swarming motility has also been described in other bacteria, such as scr mutants of Vibrio parahaemolyticus. The scrABC operon in which scrC encodes for a GGDEF protein regulates lateral flagellar expression, swarming and capsular polysaccharide production (Boles and McCarter, 2002; Guvener and McCarter, 2003). Furthermore, the RocS protein containing a GGDEF domain regulates the rugose phenotype, EPS, biofilm formation and motility in Vibrio cholerae (Rashid et al., 2003). Second, the number of GGDEF proteins in some bacteria is remarkably high (41 in V. cholerae, 33 in Pseudomonas aeruginosa) and it is therefore reasonable to speculate that they will not only be involved in the regulation of a singular metabolic pathway. Third, for the members of the family that have been described in other bacteria, it has been shown that they are involved in many different processes such as cell aggregation and biofilm formation (Jones et al., 1999; D’Argenio et al., 2002; Bomchil et al., 2003; Rashid et al., 2003), cellulose and exopolysaccharide biosynthesis (Tal et al., 1998; Ausmees et al., 1999; 2001; Boles and McCarter, 2002; , 2002; Guvener and McCarter, 2003; Rashid et al., 2003; Spiers et al., 2003), pole development of Caulobacter crescentus (Aldridge and Jenal, 1999; Paul et al., 2004) and survival and replication in humans (Kim et al., 2003). Fourth, the GGDEF domain is usually coupled with a variety of sensory domains in a modular fashion suggesting a role in converting different signals from the extracellular environment in the diffusible molecule cyclic di-GMP that might affect the expression of downstream genes (for a review, see Galperin et al., 2001; Jenal, 2004). For example, PleD is a multidomain protein with two N-terminal phosphate receiver modules and a C-terminal GGDEF domain. In a recent article, Paul et al. (2004) demonstrated that the activated form of PleD, PleD-P, is specifically sequestered to one pole of the cell, where it might locally increase the levels of cyclic di-GMP to control pole development of C. crescentus. In the case of S. typhimurium, the GGDEF domain is coupled in different proteins with a variety of sensor domains such as MASE1, MASE2, MHYT or signalling domains (HAMP, PAS), suggesting that each protein responds to a different environmental signal. Finally, only the mutation of adrA and gcpA affected cellulose biosynthesis and biofilm formation under the conditions tested whereas most of the proteins contributed to these processes when they were overproduced.
The redundancy of proteins with a similar enzymatic activity could always mask the functionality of their counterparts, hindering the real contribution of each individual protein. In order to solve this problem, we are producing a set of recombinant S. typhimurium strains in which all GGDEF proteins except for one are going to be deleted. Analysis of the set of strains harbouring a unique functional GGDEF protein could help to determine the contribution of each GGDEF protein to the global regulatory network of the bacteria and specifically to the cellulose biosynthesis and biofilm formation process.
Among all the GGDEF proteins analysed, only the phenotypes shown by gcpE mutation and complementation experiments might correspond to a phosphodiesterase activity. Overexpression of gcpE abolished cellulose synthesis in LB and a non-polar mutation of the gene increased the brightness on calcofluor plates suggesting a higher accumulation of cellulose. These results suggest that GcpE may act as a phosphodiesterase that catalysed the degradation of c-di-GMP to 5′-GMP. GcpE contains an EAL domain, which has been associated with a phosphodiesterase activity. However, GcpC and GcpF also contain an EAL domain and the phenotypes displayed by the mutant and complemented strains do not correspond to the ones displayed by GcpE. We are still far from understanding how GGDEF and EAL domains interact. It is not clear why sometimes both domains are required to exist simultaneously in the same protein and also why sometimes the activity of one domain prevails more than the other. Therefore, further studies are required to understand the context in which EAL-associated phosphodiesterase activity prevails the diguanylate cyclase activity of the GGDEF domain. It is worth noting that overexpression of gcpE also inhibited the synthesis of the polysaccharide compound that connects the adrA mutant (see Figs 5A, 6–8) or the bcs operon mutants (Solano et al., 2002), indicating a common regulation for the production of both exopolysaccharides.
The non-availability of commercial c-di-GMP has hindered biochemical studies to determine the enzymatic activity and optimal reaction conditions of the members of the GGDEF protein family. Although a genetic approach can still contribute to clarifying the function of these proteins, only the inclusion of biochemical approaches in the study of the role of GGDEF proteins in the regulation of the levels of c-di-GMP will clarify the importance of this molecule as a secondary signal transducer in bacteria.
Bacterial strains, plasmids and culture conditions
The most relevant bacterial strains and plasmids used and constructed in this study are listed in Table 2. For routine culture and cloning, bacteria were grown at 37°C in Luria–Bertani Lennox (LB) broth or on LB agar (Pronadisa) with appropriate antibiotics at the following concentrations: kanamycin (Km), 50 µg ml−1; chloramphenicol (Cm), 20 µg ml−1, ampicillin (Am), 100 µg ml−1, tetracycline (Tet), 10 µg ml−1, apramycin (Apr), 60 µg ml−1. The genomic bank of chromosomal DNA of S. typhimurium 14028 kindly provided by Ana I. Prieto was constructed by cloning DNA fragments digested with Sau3AI to give an average fragment size of 8 kb into pBR328 that had been digested with BamHI and treated with alkaline phosphatase.
pBR328::STM1987 from 14028 containing E479G and E480S amino acid substitutions
pBR328 containing STM4551 from 14028
pBR328 containing yegE from 14028
pBR328 containing yeaJ from 14028
pBR328 containing yci R from 14028
pBR328 containing STM3388 from 14028
pBR328 containing yfi N from 14028
pBR328 containing mlrA from SL1344
pBR328 containing csgD from 14028
General molecular techniques
Routine DNA manipulations were performed using standard procedures (Ausubel et al., 1990) unless otherwise stated. Plasmid DNA from E. coli and Salmonella was purified with a Quantum Prep Plasmid kit (Bio-Rad). Plasmids were transformed into E. coli and Salmonella by electroporation. Transformants carrying the Red helper plasmid were made electrocompetent with the following protocol. Cells were grown overnight in LB broth Am at 30°C, and then used to inoculate 500 ml of LB broth Am that were incubated with aeration at 30°C to an OD600 of 0.2. Then, l-arabinose (Sigma) was added to a final concentration of 0.08% and incubation continued until the OD600 reached 0.7. The suspension was cooled down on ice for 15 min and washed twice with the same volume and then once with 40 ml of ice-cold 10% glycerol. Cells were finally resuspended in 1.5 ml of ice-cold 10% glycerol. Restriction enzymes were purchased from Boehringer Mannheim and used according to the manufacturer's instructions. Oligonucleotides were obtained from Life Technologies (TableS2). Phage P22 HT105/1 int-201 (Schmieger, 1972) was used to carry out transductions between strains according to recommended protocols (Maloy et al., 1996).
Phenotypic assays for biofilm formation
Screening of the DNA library clones that bestowed strain SL1344 with the capacity of cellulose production was carried out by qualitatively assessing the level of calcofluor (Fluorescent brightener 28; Sigma) binding of colonies grown on LB agar supplemented with calcofluor 200 µg ml−1, at room temperature for 48 h. Fluorescence of the cells was observed under a 366 nm UV light source and compared with the wild-type strain. The biofilm-forming assay in ATM was carried out as previously described and visualized as a ring of cells adhered to the glass wall at the air–liquid interface after 4 h of incubation at 37°C under strong shaking conditions (Solano et al., 2001). The biofilm formed in standing LB broth was visualized after 96 h of incubation at room temperature as a floating pellicle at the air–broth interface that totally blocked the surface of the culture and could not be dispersed by shaking.
DNA sequencing was carried out by the double-strand dideoxy-chain termination method by using an ABI Prism 310 automatic DNA sequencer (PE Applied Biosystems) according to manufacturer instructions. Sequence homologies to the genes in the GenBank database were determined by using the blast algorithm of the Nacional Center for Biotechnology Information at the National Library of Medicine.
One-step inactivation of chromosomal genes
For disruption of the mlrA, adrA, gcpA, gcpB, gcpC, gcpD, gcpE, gcpF and gcpG genes in S. typhimurium 14028, PCR-generated linear DNA fragments were used as previously described (Solano et al., 2001). Briefly, the Red helper plasmid pKOBEGA (Chaveroche et al., 2000) was introduced into S. typhimurium 14028 by heat shock. Transformants were selected on LB agar Am after incubation for 24 h at 30°C. One transformant carrying the Red helper plasmid was made electrocompetent as described above. A selectable antibiotic resistance gene was generated by PCR from chromosomal DNA isolated from E. coli MC4100 (Km) and MC4100 F′tet ΔtraD::aac (Apr), using primer pairs of 80 nt-long primers that included 60 nt homology extensions for the targeted locus and 20 nt priming sequences for the kanamycin or apramycin resistance gene as template. Primer pairs used for disruption of the mlrA, gcpA, adrA, gcpB, gcpC, gcpD, gcpE, gcpF and gcpG genes are described in TableS2. Electroporation (25 µF, 200 Ω, 2.5 kV) was carried out according to the manufacturer's instructions (Bio-Rad) by using 50 µl of cells and 1–5 µg of purified and dialysed (Nitrocellulose filters 0.025 µm; Millipore) PCR product. Shocked cells were added to 1 ml of LB broth, incubated for 1 h at 30°C, and then spread onto LB agar Km/Apr to select KmR/AprR transformants after incubation at 37°C for 24 h. Mutants were then grown in LB broth Km/Apr at 43°C for 24 h and then incubated overnight on LB agar Km/Apr, Am at 30°C to test for loss of the helper plasmid.
The plasmid pI (Fig. 2) was digested with EcoRV and religated to give pIa containing dsrA, yedP and STM1987 genes. STM1987, STM1988-yedI and yedA genes (Fig. 2) were amplified on plasmid pI (Table 2) with high-fidelity thermophilic DNA polymerase (Dynazyme Ext, Finnzymes) using the primers described in TableS2. The STM2156A and yehS, yehTU, mlrA and STM2161 were amplified on plasmid pII with the same DNA polymerase using primers described in TableS2. The PCR-amplified fragments were cloned in pGEM®-T Easy (Promega), sequenced and digested with EcoRI or PstI to clone them into pBR328 to give pIb, pIc, pId, pIIa, pIIb, pIIc and pIId (Table 2). The mlrA gene was amplified on chromosomal DNA isolated from S. typhimurium 1344 with the same DNA polymerase using primers yehVExt.Fw/yehVExt.Rv (TableS2). The PCR-amplified fragment was cloned in pGEM®-T Easy (Promega), sequenced and digested with EcoRI to clone it into pBR328 to give pBR328::mlrA(1344). All these plasmids were transformed by electroporation into S. typhimurium SL1344 and the complemented strains were assessed for their ability to form a biofilm in ATM and LB as well as for fluorescence under UV light on LB agar supplemented with calcofluor. The adrA, gcpB, gcpD, gcpE, gcpG, gcpC, gcpF and csgD genes containing their own promoters were amplified on chromosomal DNA isolated from S. typhimurium 14028 with high-fidelity thermophilic DNA polymerase (Dynazyme Ext, Finnzymes) using primers described in TableS2. The PCR-amplified fragments were cloned in pGEM®-T Easy (Promega), sequenced and cloned into pBR328 to give pBR328::adrA, pBR328::gcpB, pBR328::gcpD, pBR328::gcpE, pBR328::gcpG, pBR328::gcpC, pBR328::gcpF, pBR328::csgD (Table 2). Plasmids were transformed by electroporation into S. typhimurium 14028 ΔSTM1987-Km and S. typhimurium 14028 ΔadrA-Km. pBR328::adrA and pBR328::csgD were transformed into S. typhimurium SL1344 (Table 2). The complemented strains were assessed for their ability to form a biofilm in ATM and LB as well as for fluorescence under UV light on LB agar supplemented with calcofluor. RT-PCR experiments were performed to confirm the increased transcriptional level of each gene in the complemented strain with respect to the wild-type strain.
GG[DE]EF motif mutagenesis of the GGDEF domain
Two fragments of the STM1987 gene were amplified on chromosomal DNA isolated from S. typhimurium 14028 with high-fidelity thermophilic DNA polymerase (Dynazyme Ext, Finnzymes) using primers described in TableS2. The PCR-amplified fragments were cloned in pGEM®-T Easy (Promega) and digested with EcoRI and BamHI. Two fragments were ligated and cloned in pGEM®-T Easy (Promega) and digested with EcoRI to be cloned into pBR328 to give pBR328::STM1987-GGGSF. Thus, we obtained a STM1987-cloned gene (STM1987-GGGSF) that contained the following amino acid substitutions: E479G and E480S. This construction was sequenced and transformed by electroporation into S. typhimurium 14028 ΔSTM1987-Km.
Total RNA of the wild-type strains S. typhimurium 14028, SL1344 and the complemented strains was isolated after 24, 48 and 72 h of incubation in LB under biofilm-forming conditions and after 2 h of incubation in ATM biofilm-forming conditions. Total RNA was obtained using the Fast RNA-Blue kit (Bio 101) according to the manufacturer's instructions. Two micrograms of each RNA were subjected in duplicate to DNAse I (Invitrogen) treatment for 30 min at 37°C. The enzyme was inactivated at 65°C in the presence of EDTA 0.25 mM during 10 min. The RNA samples were reverse transcribed in the presence or absence of the enzyme M-MLV Reverse Transcriptase (Promega), respectively, in order to verify the abscence of contaminating genomic DNA. All preparations were purified using CentriSep spin columns (Princeton separations). One-twentieth of each reaction was used in triplicate for real-time semi-quantitative PCR using the SYBR Green PCR Master Mix (Applied Biosystems) in the ABI Prism 7900 HT (Applied Biosystems). The transcripts of mlrA, csgD, adrA, gcpA, gcpB, gcpC, gcpD, gcpE, gcpF and gcpG were amplified using primers described in TableS2. The gyrB gene was selected as the endogenous control of the experiment as it is constitutively expressed (Wolz et al., 2002). The primer concentration was previously optimized and standard curves were obtained for every gene in order to verify that the amplification efficiency was similar so that the formula 2–Δct could be applied. To monitor the specificity, final PCR products were analysed by melting curves. Only samples with no gyrB amplification of the minus reverse transcriptase aliquot were considered in the study. The amount of RNA was expressed as the n-fold difference relative to the control gene (2–ΔCt, where ΔCt represents the difference in threshold cycle between the target and control genes). The final value is represented as the mean and standard deviation of five independent experiments.
The data corresponding to gene expression were compared using the Kruskal–Wallis and the Mann–Whitney tests. All the tests were two sided and the significance level was 5%. The spss 11.0 statistical package was used.
We express our gratitude to Ana I. Prieto for providing us with the genomic bank of S. typhimurium 14028. Begoña García and Cristina Latasa are predoctoral fellows from the Departamento de Educación y Cultura and Departamento de Salud del Gobierno de Navarra respectively. This work was supported by the BIO2002-04542-C02 Grant from the Comisión Interministerial de Ciencia y Tecnología, ‘Beca Ortiz de Landazuri’ Award from the Departamento de Salud and Grant 17/2004 from the Departamento de Educación y Cultura del Gobierno de Navarra, Spain.